Career Development Plan and Institutional Resources Summary. The career development plan consists of five years of focused instruction in epidemiology and health services research. Through a combination of course work and individual tutorials, enhanced research skills will be developed in three core areas: health economics and cost analysis; advanced methods in statistical modeling; and introduction to the use of administrative and national survey data. These new skills will provide a foundation for an independent research career in dementia, health services research focused on determinants of health care utilization, costs and response to interventions in naturally occurring populations. Michigan State University has a strong tradition in both epidemiology and health serviceS research. This environment will provide frequent opportunities for interaction will colleagues with similar research interests for career guidance, critique of ongoing research, and development of new research ideas. Research Plan Summary. Alzheimer's disease (AD) imposes a great economic burden on patients, their families and society. New treatments for AD are being developed to modify clinical features of the disease, but the relationships between these clinical features and direct health care costs are not known. We will build a data set for investigating determinants of direct heath care costs in an AD population. The Behavioral Model of Health Services Use will provide a conceptual framework for data collection and analysis. We will measure direct health care costs in a well-characterized cohort of 300 probable AD patients over four years, by repeated survey measurements and linkage to HCFA administrative data. We will examine the relationships between direct costs and six "need" factors hypothesized to be important determinants of direct, health care costs in AD. These six "need" factors (ie, severity of the AD patient's cognitive, psychiatric, and motor impairment, co-morbid medical conditions, "dependency" and caregiver "perceived burden") will be measured with validated, clinical measurement instruments. We will use multiple regression analysis techniques, including path analysis, structural equation modeling, and general longitudinal modeling to test study hypotheses. This project will develop explanatory and predictive statistical models of direct health care costs in AD patients, that can help focus and evaluate future interventions in this disease.